LLaMA-Mesh Text Generator

A web-based text generation application powered by the LLaMA-Mesh model, built with Flask and Transformers.

📖 Overview

This project provides a clean, user-friendly web interface for generating text using the LLaMA-Mesh language model. Users can input prompts and control the length of generated text through an intuitive web form.

✨ Features

• Web Interface: Clean, responsive design using Tailwind CSS
• Text Generation: Powered by the LLaMA-Mesh model via Transformers
• Customizable Output: Adjustable text length (10-1000 characters)
• Real-time Processing: Asynchronous text generation with loading states
• Error Handling: Graceful error display and recovery

🛠️ Technology Stack

• Backend: Python Flask
• ML Framework: PyTorch + Transformers
• Model: LLaMA-Mesh (Zhengyi/LLaMA-Mesh)
• Frontend: HTML, JavaScript, Tailwind CSS
• Package Management: pip + virtual environment

📋 Requirements

System Requirements

• RAM: 32GB+ recommended (model is ~15GB)
• Storage: 20GB+ free space
• GPU: CUDA-compatible GPU recommended (optional)
• Python: 3.8 or higher

Dependencies

• Flask >= 3.1.0
• torch >= 2.2.0
• transformers >= 4.30.0
• Additional dependencies in requirements.txt

🚀 Installation

1. Clone the Repository

git clone <your-repository-url>
cd capstone

2. Create Virtual Environment

python -m venv .venv

3. Activate Virtual Environment

Windows:

.venv\Scripts\activate

macOS/Linux:

source .venv/bin/activate

4. Install Dependencies

pip install -r requirements.txt

🎯 Usage

Starting the Application

python app.py

The application will start on http://localhost:5000

Using the Web Interface

1. Open your browser and navigate to http://localhost:5000
2. Enter your text prompt in the input field
3. Adjust the maximum length slider (10-1000 characters)
4. Click "Generate" to create text
5. View the generated output below

API Endpoint

You can also use the API directly:

curl -X POST http://localhost:5000/generate \
  -H "Content-Type: application/x-www-form-urlencoded" \
  -d "input_text=Your prompt here&max_length=100"

📁 Project Structure

capstone/
├── app.py              # Main Flask application
├── index.html          # Web interface template
├── requirements.txt    # Python dependencies
├── .gitignore         # Git ignore rules
├── .venv/             # Virtual environment (ignored)
└── README.md          # This file

⚙️ Configuration

Model Parameters

You can modify these parameters in app.py:

• temperature: Controls randomness (0.1-2.0)
• top_p: Nucleus sampling parameter (0.1-1.0)
• max_length: Maximum output length
• num_return_sequences: Number of sequences to generate

Flask Configuration

• Debug Mode: Enabled by default (debug=True)
• Host: localhost (127.0.0.1)
• Port: 5000

🔧 Troubleshooting

Common Issues

1. Memory Error / Segmentation Fault

Problem: Model too large for available RAM Solutions:

• Use a machine with more RAM (32GB+)
• Use a smaller model variant
• Enable model offloading to disk

2. Import Error (Werkzeug/Flask)

Problem: Version compatibility issues Solution:

pip install --upgrade flask werkzeug

3. CUDA Out of Memory

Problem: GPU memory insufficient Solutions:

• Use CPU-only inference: torch.device('cpu')
• Reduce batch size
• Use gradient checkpointing

4. Slow Model Loading

Problem: Large model download/loading time Solutions:

• Models are cached after first download
• Use faster internet connection for initial download
• Consider using model quantization

Performance Optimization

1. GPU Acceleration: Ensure CUDA is properly installed
2. Model Caching: Models are cached in ~/.cache/huggingface/
3. Memory Management: Use torch.no_grad() for inference
4. Batch Processing: Process multiple requests efficiently

🤝 Contributing

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

📝 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• LLaMA-Mesh Model by Zhengyi
• Transformers by Hugging Face
• Flask web framework
• Tailwind CSS for styling

📞 Support

If you encounter any issues or have questions:

1. Check the Issues page
2. Review the troubleshooting section above
3. Create a new issue with detailed information

🔮 Future Enhancements

• Model selection dropdown
• Batch text generation
• Export/save functionality
• User authentication
• API rate limiting
• Docker containerization
• Cloud deployment guides

---

Note: This application requires significant computational resources due to the large language model. Ensure your system meets the minimum requirements before installation.